Talk by Pentti Kanerva at the Redwood Center for Theoretical Neuroscience lab meeting.
ABSTRACT Traditional computing is deterministic. It is based on bits and assumes that the bits compute perfectly. However, (near) perfection at high speeds consumes large amounts of energy and limits the scaling-down of the underlying circuit elements. Contrast this with the brain's powers of perception and learning. They go far beyond what computers can do, are accomplished with little energy by slow and imprecise neurons wired together according to a general plan, with many details left to chance. Can we make sense of that kind of computing and build systems on the same principles? To quote a campaign slogan of yesteryear, "Yes We Can."